Open sstemann opened 3 years ago
We use Disease Ontology and the Node Normalizer does not return an equivalent for "MONDO:0007114"
TRAPI validation is failing on the type of "Status" that we're returning. This is a known issue and is set to be fixed along with other TRAPI 1.1 work.
Is there a DOID for Angel-shaped phalango-epiphyseal dysplasia? Mondo doesn't reference it...
Sorry @cbizon , I didn't meant to suggest that this was a problem with the NodeNormalizer. No, I don't think there is a DOID for this specific condition, which is the true cause of the problem.
No worries, there are certainly problems - I just wasn't sure if this was one of them :)
Text Mining Provider should return results for this type of query in the future, but we are still working on extracting this relation from text, so the data is not currently in our KG.
Sorry @cbizon , I didn't meant to suggest that this was a problem with the NodeNormalizer. No, I don't think there is a DOID for this specific condition, which is the true cause of the problem.
@brettasmi I tried the same query with DOID:9252 (inherited amino acid metabolic disorder) parent PK: d6acd0f6-58b9-4f60-a156-c2a5b6281329
Improve returns this message: Could not find any supported predicates for subject category: ['biolink:PhenotypicFeature']; returning empty message...
@webyrd when I was re-running the query for a DOID, I used DOID:9252 and Unsecret returned 0, I switched to the corresponding MONDO:0004736 and Unsecret returned 1204 results - was wondering if you could look in to that as well?
@brettasmi I tried the same query with DOID:9252 (inherited amino acid metabolic disorder) parent PK: d6acd0f6-58b9-4f60-a156-c2a5b6281329
This works if you switch the subject and object as discussed on the call.
I just talked with @gregr about the 174 results returned by Unsecret Agent, and why the predicates in the returned answers differ from the predicate in the query. When Unsecret can't find any matching edges with the predicate in the query, we expand the query to allow for all predicates.
We are working on lightweight reasoning/query expansion, to expand the queries according to the Biolink model's predicate ontology, which should help us return better results.
Since some of the local KGs we have imported are not yet Biolink compliant, we may be missing data that exist in those KGs when given a Biolink predicate. Also, this is why we return a non-Biolink predicate in the results.
After reversing subject and object, BTE returns 15 edges in the result.
$ curl --silent --location --request POST "https://api.bte.ncats.io/v1/query/" --header "Content-Type: application/json" --data-raw "{
\"message\": {
\"query_graph\": {
\"edges\": {
\"e01\": {
\"object\": \"n1\",
\"subject\": \"n0\",
\"predicate\": \"biolink:has_phenotype\"
}
},
\"nodes\": {
\"n0\": {
\"id\": \"MONDO:0007114\",
\"category\": \"biolink:Disease\"
},
\"n1\": {
\"category\": \"biolink:PhenotypicFeature\"
}
}
}
}
}" | jq '.message.knowledge_graph.edges | length'
15
CAM-KP doesn't contain any biolink:Disease
or biolink:PhenotypicFeature
, so 0 results is expected.
I ran the updated query below through the ARS, PK: e7a9ba48-45b1-42f3-9fb7-779eeb0f691a
The updated query could be something like this (it's missing the workflow section but the rest is TRAPI v1.2):
{
"message": {
"query_graph": {
"nodes": {
"n0": {
"ids": ["MONDO:0007114"],
"categories":["biolink:Disease"]
},
"n1": {
"categories": ["biolink:PhenotypicFeature"]
}
},
"edges": {
"e0": {
"subject": "n0",
"object": "n1",
"predicates":["biolink:has_phenotype"]
}
}
}
}
}
Query: RarePhenotype PK: 15480a55-4db9-4fde-990c-97b08c06d5e4 MONDO:0007114 Results Tracking Sheet
Responses From: